In November, 1989, soil samples were collected from each of the five grass monocultures in each of the 11 plots. These 55 samples were composites formed by combining a single core (2.2cm in diameter X 15cm deep) from the center of each replicate monoculture of each of the grass species within each of the large garden plots. There were two replicate monocultures of Andropogon and three replicate monocultures of the other four species within each of the large plots. Soils were placed in Whirlpack bags and stored at 4 degrees C until they were processed. Composite soil samples were thoroughly mixed and spores were extracted from 25g of air-dried soil using a modification of McKenney and Lindsey's (1987) technique. We used a smaller mesh screen (25um) than that recommended by McKenney and Lindsey (38um) to minimize loss of the smallest spores. Spores were spread evenly onto a gridded membrane filter and all of the spores within 27% of the area of the filter paper were removed using a dissecting microscope (50x) and a fine forceps. These spores were mounted in both polyvinyl-alcohol and a 1:1 solution of polyvinyl-alcohol plus Melzer's solution to make permanent slides. Slide were examined using a microscope (100 - lOOOX) and spores were identified to species based on wall structure (Schenck and Perez, 1990). Voucher specimens of each of the species we identified are available from the plant pathology herbarium at the University of Minnesota.

The relative density (%) of each of the species in each composite sample was calculated as: (ni /Nt) x 100; where n i = number of spores from the ith species and Nt spores examined in the sample (Nt averaged 935 total number of spores). Total spore counts of each species were estimated by multiplying relative spore densities by the total count of spores on the filter paper. Diversity of VAM fungal communities were calculated by the Shannon-Wiener index (Krebs, 1985). Percent similarity of VAM fungal communities by host plant and by soil mixture were computed and cluster analysis performed using MVSP Plus, version 2.0 (Kovach, 1990).

It is important to note that relative spore density does not necessarily reflect the functional importance of individual species within the VAM fungal community because some species may be prolific sporulators even when they are in relatively low abundance within plant roots. Spores of Glomus aggregatum and Glomus leptotichum were significantly more abundant than spores of all other species across all combinations of host plants and soil mixtures. Spore densities of these two species were omitted from the similarity analysis because their over dominance obscured the distribution patterns of less abundant but potentially equally important species in the VAM fungal community.